Frequently asked questions
Short, honest answers. If something is missing, open an issue on GitHub.
β 01 Which browsers work?
Anything Chromium-based: Chrome, Edge, Brave, Arc, Vivaldi, Opera. Safari does not yet support getDisplayMedia with audio reliably. Firefox can capture screen, but audio support is limited β Chromium is the sweet spot.
β 02 Why does my Mac only capture tab audio, not Zoom or Spotify?
macOS browsers can only capture audio from inside a browser tab. To capture system-wide audio (Zoom, Spotify, Skype, etc.), install BlackHole (free, open-source virtual audio device) and route the app's audio through it. On Windows and Linux this 'just works' via screen share.
β 03 How do I get a Groq API key?
Go to console.groq.com, sign up (Google login works), then visit /keys and create a key. It takes about 60 seconds. The free tier gives you about 8 hours of transcription per day per key.
β 04 Is my audio sent to your server?
No. There is no server. The app is a static HTML/JS bundle on Cloudflare Pages. Audio chunks go from your browser straight to Groq's API. Translated text goes from your browser straight to Google / Microsoft / DeepL / MyMemory. We have no logs, no analytics on the /app page, no opt-out toggle because there is nothing to opt out of.
β 05 How private is it really?
As private as you trust your browser and Groq. Your Groq key sits in localStorage. Your transcripts sit in localStorage. Nothing is stored on our infrastructure because we don't have any infrastructure beyond serving static files.
β 06 What languages does it support?
Speech recognition: every language Whisper Large v3 supports (~100). Translation between any of: English, Vietnamese, Chinese, Japanese, Korean, French, Spanish, German, Russian, Arabic, Hindi, Thai, Indonesian, Portuguese, plus more via Google fallback.
β 07 How accurate is the translation?
Google's free endpoint is solid for everyday conversation. For technical, legal, or medical content, plug in a DeepL API key (free tier 500k chars/month) β the difference is noticeable.
β 08 Can I export the transcript?
Yes. The transcript panel has Copy, Download .txt, and Download .srt buttons. The .srt file can be loaded as a subtitle track in VLC, mpv, or YouTube.
β 09 Will the desktop app come back?
A separate Tauri desktop build exists for power users who want offline whisper.cpp + GPU acceleration. The web version covers 95% of use cases without any install.
β 10 Why isn't there a paid plan?
Because there's no server to pay for. We serve static files on Cloudflare's free tier. You pay Groq nothing on the free tier, or pennies if you go beyond it. Adding a paid layer would mean adding a backend, which would defeat the privacy story.